Xiaomi Auto World Model: A Joint World Model Integrating Reconstruction and Generation for Autonomous Driving

arXiv 2026 48.6 method, system, application

TLDR

A unified world model for autonomous driving combining reconstruction (WorldRec) and generation (WorldGen) for closed-loop simulation and data synthesis.

Reasoning

The paper presents a novel joint architecture integrating feed-forward 3D scene reconstruction with efficient causal video generation, achieving synergistic gains in consistency and fidelity. Strengths include a clear two-module design and progressive training; weaknesses are the lack of explicit real-world experimental results in the abstract and potential complexity in deployment.

Read-first score

Read-first score 48.6, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 42.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

Citation impact 18%

71.9

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.71878087

Topical relevance 29%

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Methodology quality 18%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=none

Reproducibility 18%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

FrontierBridge

Rank sensitivity

Stability: volatile; rank range: 333.

Keyword Scores

world model

generative world model

video world model

world simulator

world dynamics prediction

interactive world model

model-based reinforcement learning world model

Deep Analysis

Innovations

WorldRec: a feed-forward reconstruction architecture driven by sparse scene queries that initializes structured queries in 3D space to aggregate cross-view, cross-temporal features, enforcing spatial consistency and yielding compact high-fidelity 3D Gaussian scene representations.
WorldGen: a two-stage training framework of bidirectional pretraining followed by causal fine-tuning through three progressive stages (Teacher Forcing, ODE distillation, and DMD), enabling high-quality online causal video generation in as few as 4 denoising steps.
JWM: deep integration of WorldRec and WorldGen to achieve synergistic gains in generation stability, cross-frame consistency, and visual fidelity.

Methodology

The paper proposes a unified system with two modules: WorldRec for world representation using sparse scene queries in 3D space to aggregate cross-view and cross-temporal features, and WorldGen for world generation using a two-stage training framework (bidirectional pretraining then causal fine-tuning with Teacher Forcing, ODE distillation, and DMD) for efficient video generation. The Joint World Model (JWM) integrates both modules for synergistic improvements.

Key Results

The JWM achieves synergistic gains in generation stability, cross-frame consistency, and visual fidelity, providing a solid foundation for closed-loop simulation, data synthesis, and end-to-end training in autonomous driving.

Resources

Paper